Yasmin, as you already noticed, you will need to perform a multiple regression analysis. The regression equation looks like this:

Intention = b0 + b1*Attitude + b2*Subj. Norms + b3*PBC + e

In order to determine the importance of the predictors, use the standardized solution in SPSS. The parameter values of beta1, beta2, and beta3 (the standardized parameters) are usually interpreted if you are not interested in the original metric.

Does this help?

Regards, Karin

Dear Prof. Engel, 

this is definitely helpful, thank you very much. :)

Alternatively, you could ask people what they intend to do. The predictors do not equate with intention. They may think the action is desirable (Attitude), they may think other people do this (norm) but they may have zero intention of doing anything themselves.  Alternatively, people may intend to do something to a certain extent.  However, if they are antipathetic toward it, yet, believe others expect it of them, they may express some modest intention to do the act. You can have a low through high level of intention to act, but at teh end of the day around 40% of those intending to act actually do so.  How well attitude and norms predicts intention is also somewhat available, depending on other factors like culture (not the same as norm - some people show more norm-sensitivity than others.)

Hi Alex, 

thank you very much for the detailed answer! :)

I measured the constructs with different items on unipolar Likert-scales, except for the Attitude and the Intention which were measured through semantic differentials. I will do the regression now and let SPSS calculate the betas etc, and see how it goes. 

Thanks a lot! 

Hopefully you have determined your hypothesis a priori and are testing one specific prediction.  If you're just looking for associations between any of the variables on the off-chance they are related, then that is not a good idea.

What I understand from Ajzen statement is that the Three Component ( Attitude, Subjective Norm and Perceived Behavioral Control) will explain direct effect (regression weight) to Intention. The Intention can not be measured because its known as Latent Variable or Construct which dependent on the Three Component (ATT, SN and PBC). Some use term as Exogenous (Independent) variable and Endogenous (Dependent) variable .

If, you prefer to use other Dependent (Behavioral), then it should consider Intention as Dependent 1st and Behavioral as Dependent 2nd. In that case, you should applying Hierarchical Multiple Regression or Structural Equation Model.

Hello Richard, yes, I have some hypotheses. :) But out of curiousitiy - why would it be a bad idea to look for associations/correlations? 

That is fishing.  When you fish for associations you will always find some.  There will always be associations that emerge purely by chance.  Ideally. you should be able to predict the associations that will emerge based on the theory.  If you do not find these, and you have a good randomized sample of sufficient size, then it is a good bet that the theory is failing in the predictions it makes. A major problem with so many psychology papers in the literature is that they are not reproducible.Better theory is needed  This suggests that the associations reported are spurious.  and this is more likely to happen if a "fishing" approach to the data is used.  Ideally, a good theory makes predictions.  TPB studies generally tend to account for around 35-40% of the observed variance between intention and outcome. The so-called "intention-behaviour gap" remains a big problem and a modified TPB can yield better prediction, around 50-60%. But that means around half of the observed variance is not accounted for by model variables. 

It's hard to overestimate the importance of theory for advancing scientific evidence for psychology (or anything else). In the 40 years I've been doing it, it has moved very little in terms of theory, and I attribute this largely to a tendency to publish associations rather than building solid evidence tests of theory.  If theory predicts well, then it is likely to be a good approximation of reality - if it doesn't, it isn't.  That means it's wrong, even if there is a statistical association in the data. I think we are all a little guilty of retrospectively writing a paper as if we planned it as reported, with absolutely no post hoc modifications. 

So that's why I say don't go looking for things, but predict them first and then see if observation matches the prediction.  If it does. That's great. If it doesn't then improve the model and reiterate.

That is fishing.  When you fish for associations you will always find some.  There will always be associations that emerge purely by chance.  Ideally. you should be able to predict the associations that will emerge based on the theory.  If you do not find these, and you have a good randomized sample of sufficient size, then it is a good bet that the theory is failing in the predictions it makes. A major problem with so many psychology papers in the literature is that they are not reproducible.Better theory is needed  This suggests that the associations reported are spurious.  and this is more likely to happen if a "fishing" approach to the data is used.  Ideally, a good theory makes predictions.  TPB studies generally tend to account for around 35-40% of the observed variance between intention and outcome. The so-called "intention-behaviour gap" remains a big problem and a modified TPB can yield better prediction, around 50-60%. But that means around half of the observed variance is not accounted for by model variables. 

It's hard to overestimate the importance of theory for advancing scientific evidence for psychology (or anything else). In the 40 years I've been doing it, it has moved very little in terms of theory, and I attribute this largely to a tendency to publish associations rather than building solid evidence tests of theory.  If theory predicts well, then it is likely to be a good approximation of reality - if it doesn't, it isn't.  That means it's wrong, even if there is a statistical association in the data. I think we are all a little guilty of retrospectively writing a paper as if we planned it as reported, with absolutely no post hoc modifications. 

So that's why I say don't go looking for things, but predict them first and then see if observation matches the prediction.  If it does. That's great. If it doesn't then improve the model and reiterate.

That is fishing.  When you fish for associations you will always find some.  There will always be associations that emerge purely by chance.  Ideally. you should be able to predict the associations that will emerge based on the theory.  If you do not find these, and you have a good randomized sample of sufficient size, then it is a good bet that the theory is failing in the predictions it makes. A major problem with so many psychology papers in the literature is that they are not reproducible.Better theory is needed  This suggests that the associations reported are spurious.  and this is more likely to happen if a "fishing" approach to the data is used.  Ideally, a good theory makes predictions.  TPB studies generally tend to account for around 35-40% of the observed variance between intention and outcome. The so-called "intention-behaviour gap" remains a big problem and a modified TPB can yield better prediction, around 50-60%. But that means around half of the observed variance is not accounted for by model variables. 

It's hard to overestimate the importance of theory for advancing scientific evidence for psychology (or anything else). In the 40 years I've been doing it, it has moved very little in terms of theory, and I attribute this largely to a tendency to publish associations rather than building solid evidence tests of theory.  If theory predicts well, then it is likely to be a good approximation of reality - if it doesn't, it isn't.  That means it's wrong, even if there is a statistical association in the data. I think we are all a little guilty of retrospectively writing a paper as if we planned it as reported, with absolutely no post hoc modifications. 

So that's why I say don't go looking for things, but predict them first and then see if observation matches the prediction.  If it does. That's great. If it doesn't then improve the model and reiterate.